3results about How to "Guaranteed recognition speed" patented technology

The invention provides a solid wood panel defect identification method based on deep learning, relates to the technical field of solid wood panel defect detection, and solves the problems that wood defect detection and classification mainly depend on manual selection, the detection method has relatively large subjective errors and the like, human eyes are fatigued, and the detection efficiency is high. The technical problems that the requirement for detection quality grading of industrial production cannot be met, meanwhile, the labor cost is higher and higher, so that the wood processing cost is higher and higher, the standard cannot be unified, and the actual detection and classification are difficult to guarantee enough precision are solved. According to the technical scheme, a binary classification network is introduced to the front end of a traditional defect detection algorithm, and classification of defective wood and defect-free wood is completed in advance through the network; according to the method, the waste of defect detection algorithm resources is effectively reduced, and the efficiency of identifying the defects of the solid wood panel is improved.

The application provides a kind of dynamic gesture recognition method and system based on data glove, it is related to signal processing and pattern recognition field, the method includes the following steps: S1: the sensing data information of different gesture actions when subject wears data glove is acquired, the sampling frequency of host computer program is adjusted to 30Hz, different gesture actions are collected under dynamic gesture acquisition mode 1-2s, and record and save;S2: the training of gesture recognition model is carried out to the sensing data information collected, MS-1D-CNN neural network is used for training, and the data of three modes are extracted and fused, to change the uniform length of sample to 60 after resampling mode, and then input into CNN;S3: the starting and ending position points of gesture are analyzed by using multi-section threshold detection process.The method can accurately and real-time recognize the gesture actions when subject wears data glove.

This invention discloses a feature fusion method and a target recognition method based on this method. The feature fusion method includes a feature fusion network. The feature fusion network fuses feature maps of different sizes extracted from the backbone network, fusing feature map information from shallow networks with feature map information from deep networks. The fused feature maps are then convolved to form a detection head. This invention improves upon the one-stage target detection algorithm YOLO V3 by fusing features from shallow and deep networks. This feature fusion method can fuse features from shallow and deep networks, minimizing the loss of information from shallow networks; and can improve the accuracy of target recognition to a certain extent when performing target recognition tasks.